How Advanced Foundational Models Will Expand the Capabilities of AI (and Other Predictions for 2025)

Why read? AI predictions by non-Sam-Altman and non-AI PhD? Well, early-stage VCs spend their days working with people far smarter than us and using AI at the cutting edge of what’s possible. This allows us to see a version of what the future might hold.

In my 30-plus years in technology, I have never seen a technology – particularly the advances in and around Foundation models – have such a big impact in so many ways and so quickly. It reframed my expectations for building teams and the way companies deliver value.

With that in mind, here are my predictions for 2025.

We’re trying to contain AI

I love Mustafa Suleymans Sequence of digital species for AI because it correctly formulates the question of containment. Much of today’s AI in products suggests options for action. But increasingly, AI will act on our behalf. We have to create what we want not I want it to work.

Expect a lot of loud cage rattling in this area. If we use California’s failed SB1047 legislation as a model, Elon Musk’s pro-regulation stance and Trump’s anti-regulation stance will balance the penalties. Look for things like Anthropics Constitutional model on steroids as the industry attempts to regulate itself to ensure the safety of the base model and avoid government regulation. Additionally, expect to see more and more AI security startups helping with forms of cyber containment.

Foundation models as brand DNA

Some of you may remember the days of “Intel Inside” as a major campaign. Foundation models will give the products based on it brand power. This goes beyond the tone of voice, which, according to Claude, looks like this today:

• GPT-4: The premier, measured advisor.
• Claude: The careful, ethical advisor.
• Lama: The gruff, flexible generalist.

The tone is adjustable. How understanding and remembering are applied in a model is not the case. Models will inevitably influence how products feel.

AI for the Nobel Prize

Not entirely, but the ability to iterate on scientific knowledge will increase massively. Today, AI enables breakthroughs together with people. I agree Vinod Khosla Related: AI scientists – AI agents who are hyper-specialists – will soon regularly outperform their human colleagues in discovering insights.

Because AI scientists don’t need breaks. You have infinite patience for iteration. You have perfect data memory. And they can work around the clock. Take a look at the first glimpses of this tireless work ethic in medical research and healthcare breakthroughs. A number of start-ups are helping companies take advantage of this.

Audio interfaces are redefining “hands-free calling.”

If Alexa, Siri and Google Assistant were all version 1.0 audio interfaces, audio – with the added context of images and text – would create entirely new product experiences that recalibrate our standards.

Specialized audio only Foundation models Race with the larger foundation models to advance what is possible here. We’re already seeing great progress with Meta’s smart glasses and how far away they are from Google Glass. “Hands-free” operation will take on a whole new meaning – and expectations.

With generational AI, we are crossing the tipping point of trust

I have a good sense of direction. It only took a few traffic jams that I didn’t know about – but my maps app did – before I gave in and finally trusted that map apps knew routes better than I did.

The same will happen with generative AI output. Currently, experts are still achieving superior results. This gap will close and we will capitulate to the results of further generations of AI, provided they are at least on par and sometimes even better than the results of experts. They receive credit in the same way that we cite those we cite. Case in point: See the “Brands” section of this article.

We pay a premium for craftsmanship

Because the threshold for creation is lower – Dall-E and a number of other tools create new designs in seconds – non-digital goods with evidence of human creation will increase in value. Think of original photography or visual design, but also the ceramic mug made by an artist rather than punched out by a robot in a factory.

My adult children marvel at VCRs and love vinyl, but this is more than vintage and cool. In a world so digital, analog and physical, they become meaningful in surprising ways, especially for Gen Z and Gen Alpha.

“People were found here”

Many of my friends love Waymo because it means they don’t have to talk to anyone. But when push comes to shove, the ability to interact with a human who can express judgment that isn’t based on black box logic is important. Which approach – AI or human-forward – has more value will become the dividing line.

Expect backlash in areas like customer service, where the rush to AI will frustrate many. Humans-in-the-loop is considered preferable to fully autonomous systems. Businesses (and consumers) will encourage and value distinction.

What a year…

I end 2024 humbled and expect a lot of chaos in 2025. But I’m an optimist and trust that I’ll continue to be surrounded by really smart people who are trying to build great companies and make things better. I wish you humility, joy and much grace for the coming year.

Martina Lauchengco is a partner at Costanoa Ventures.